Journal of the Iowa Academy of Science: JIAS

Volume 107 Number Article 4

2000

Population Size Estimates for the Endangered Iowa Pleistocene Snail, Discus macclintocki Baker

Tama K. Anderson Iowa State University

Let us know how access to this document benefits ouy

Copyright © Copyright 2000 by the Iowa Academy of Science, Inc. Follow this and additional works at: https://scholarworks.uni.edu/jias

Part of the Anthropology Commons, Life Sciences Commons, Physical Sciences and Mathematics Commons, and the Science and Mathematics Education Commons

Recommended Citation Anderson, Tama K. (2000) "Population Size Estimates for the Endangered Iowa Pleistocene Snail, Discus macclintocki Baker," Journal of the Iowa Academy of Science: JIAS, 107(2), 34-41. Available at: https://scholarworks.uni.edu/jias/vol107/iss2/4

This Research is brought to you for free and open access by the Iowa Academy of Science at UNI ScholarWorks. It has been accepted for inclusion in Journal of the Iowa Academy of Science: JIAS by an authorized editor of UNI ScholarWorks. For more information, please contact [email protected]. Jour. Iowa Acad. Sci. 107(2):34-41, 2000

Population Size Estimates for the Endangered Iowa Pleistocene Snail, Discus macclintocki Baker

TAMARA K. ANDERSON 1

Ecology & Evolutionary Biology Program, Iowa State University, Ames, Iowa 50011

Discus macclintocki Baker, the Iowa Pleistocene snail, is a federally endangered species found only on algific talus slopes in northeastern Iowa and northwestern Illinois. Population size estimates for fourteen D. macclintocki populations in Iowa and Illinois ranged from 182 to 22,125 individuals. Estimates from the program CAPTURE as well as Bayesian estimation procedures gave similar results, although the Bayesian method allowed estimation of populations that could not be estimated with CAPTURE due to small sample sizes. A comparison of two methods of sampling, visual counts of quadrats and cover boards, showed that using cover boards resulted in much higher probabilities of capture and more recaptures. Snail activity was highly variable over time and space, making precise estimation difficult. Several populations are much larger than previously thought, although it is not clear whether this is due to bias in previous sampling methods or actual increases in populations since the previous estimates were made.

INDEX DESCRIPTORS: snail, population estimation, algific slopes, Discus macclintocki, endangered species, Iowa Pleistocene snail.

Discus macclintocki is a glacial relict species found only on algific to avoid mossy areas and areas covered by yews, which are common (cold-air) talus slopes in northeastern Iowa and northwestern Illinois on these slopes (Solem 1976, Frest 1981). These preferences make and nowhere else on earth. These snails are small (5 to 8 mm in for an extremely uneven distribution of individuals on the algific diameter) and either brown or olive in color. Due to their size, col­ area. The portions of the slopes with cold-air drainage are themselves oration, and the nature of their habitat, they are difficult to sample. often small and separated, so distribution over the whole slope is Baker ( 1928) first described this species from fossils, but it was not distinctly patchy. until a scientist discovered Shimek's notes in a museum regarding Studying population dynamics of terrestrial snails presents a chal­ live specimens that anyone realized the species might not be extinct lenge to malacologists. The difficulties include: extreme variability (Pilsbry 1948). Hubricht (1955) rediscovered a live population in in microhabitat (and therefore species density) from quadrat to quad­ Iowa which eventually renewed interest in the species. Afrer exten­ rat, low densities, and little movement between sampling times sive searches (Frest 1981-1987), just over thirty populations are (Goodhart 1962, Ausden 1996). Malacologists have used a variety known to exist today, all located in a unique type of habitat called of sampling methods in their studies including transects (Goodhart algific talus slopes. Largely because of the rarity of their habitat, D. 1962), grab samples (Van Es and Boag 1981, Locasciulli and Boag macclintocki was declared endangered in 1977, and is currently in­ 1987), wooden cover boards (Leonard 1959, Boag 1982, Ostlie 1992, cluded on both the Iowa (State of Iowa 1988) and the United States 1993), and, the most common method, visual searching of a grid or Endangered Species lists (U.S. Fish and Wildlife Service 1993). The quadrat (Cain and Currey 1968, Williamson et al. 1977, Cowie need for conservation and management plans for this species makes 1984, Baur and Baur 1990). However, factors such as night-time estimates of population sizes especially important. movements, very slow dispersal rates, and excellent camouflage make Algific talus slopes consist of loose, porous rock with underlying visual location difficult (Tomiyama and Nakane 1993). Multiple ap­ deposits of ice. Cold-air currents circulating through the slopes keep proaches are rarely used in the same malacological study, so com­ the talus cool. The temperatures on these slopes stay much cooler parisons of population size estimates from different methods can be and more consistent than surrounding areas throughout the summer. difficult (although see Williamson et al. 1977, Kuznik 1997, Oggier In one year, air temperatures measured at ground level on the talus et al. 1998). of one slope ranged from 3 ro 9°C year-round, whereas ground tem­ No papers have been published on the census of Discus macclintocki peratures off-slope measured -1 to + 27°C (Solem 1976). D. mac­ and most of the previous information is found only in progress re­ clintocki have a limited temperature tolerance (Frest 1981), so the ports. Population sizes for most populations of D. macdintocki were restricted temperatures on these slopes are important for the survival previously estimated by extrapolating from their densities in grab of this species. samples taken as part of studies which surveyed the snail community The cool air flows out of the talus at vents arranged haphazardly in the early 1980s (Frest 1981-1987). Due to the extreme patchiness along the slope, depending on the underlying openings in the rock. of the areas where snails are found, estimates from grab samples are The snails are found in areas with suitable temperature, moisture, difficult to replicate. In addition, grab sampling is not desirable in and vegetation, and are active at the surface where they feed and the algific talus habitats where removing talus can destabalize the mate. They prefer vegetation consisting of deciduous leaves, and tend slope, causing minor rockslides or compacting the soil and poten­ tially killing many of the snails or destroying valuable habitat (Ostlie 1 Current address: 1627 S. Summit, Newcastle, Wyoming 82701. 1993). [email protected] In 1992, Wallendorf and Clark conducted a mark-recapture study DISCUS MACCLINTOCKI POPULATION SIZES 35

of one population using basswood (Tilia americana) cover boards (Deutsch and Journel 1992). On each sampling occasion, I counted placed on the surface of the algific slopes as "trapping stations". One new individuals and recaptures. Locations of the recaptured individ­ advantage of this method is the ease with which snails can be viewed uals were noted and compared to the locations of the previous capture as they collect on the underside of the boards. The boards provide to determine the amount of movement occurring within the popu­ an attractive microhabitat for the snails, keeping them at the surface. lations. However, this method is labor intensive, since large numbers of For each slope, I measured the minimum area which included all boards must be hauled to these steep slopes. In addition, some evi­ of the sampling quadrats. Due to the geometrical arrangement of dence suggests that leaving boards on the slopes for extended periods the sampling quadrats and the unique structure of each slope, the of time reduces the snail populations under the boards (Wallendorf sampled areas are not the same on each slope. I also estimated the and Clark 1992). total algific area (i.e., potential snail habitat) on the slopes by mea­ To combat these problems, I used quadrat sampling, where work­ suring what appeared visually to be habitat of similar structure to ers visually searched each sampling plot for snails. This method elim­ that where the snails were sampled. Estimates were made of areas inated potential bias and habitat disturbance from the boards. In where direct measure was not possible (i.e. cliff faces). Dry areas or addition, I used cover boards on a subset of slopes ro determine areas with undesirable vegetation were not included as total habitat differences in the effectiveness of the two methods. available. If the snails are able to use these areas as well, the total Sampling difficulties also affect the model used to obtain a pop­ population estimates would be underestimates. On two slopes (207 ulation estimate. Snails and other invertebrates are sensitive to cli­ and 246) where I had not made measurements of total available mactic variables, so their capture probability is not constant (Ausden habitat, I used estimates by Frest (1985a, 1985b). 1996). The population estimation method used must take this into The data were compiled and analyzed using the program CAP­ account. The program CAPTURE (White et al. 1978) includes a TURE (White et al. 1978). CAPTURE uses the probability of re­ variety of models in which capture probability can vary over time, capturing marked individuals to estimate the total population size habitat, and individuals' behavior. However, large sample sizes are over the area sampled under a maximum likelihood model (White needed in order to get precise estimates (Otis et al. 1978). An al­ et al. 1978). Depending on the model chosen, the trapping proba­ ternative method of population estimation which does not require bility can be estimated as constant over the entire sampling period large numbers of recaptures is to use prior probabilities (from earlier (null model) or variable. The trapping probability may be variable captures) to calculate the probability of obtaining the observed data due to differences in resighting probability between sampling times for specific population sizes (Gazey and Staley 1986). or different behaviors of individual animals or some combination of In this study, I examined the population dynamics of Discus mac­ these (White et al. 1978). clintocki using two sampling methods, quadrat sampling and cover In contrast to Chao's model, Gazey and Staley's (1986) method of boards. In addition, two different mathematical methods for esti­ population estimation uses prior probability methods to calculate the mating population sizes are evaluated. I estimate the population size probability of obtaining the observed data for specific population and dispersal rates for fourteen populations and discuss the spatial sizes (Gazey and Staley 1986). Bayesian methods such as this can be and temporal variation associated with this species. useful with small sample sizes (Gazey and Staley 1986). I used this approach to calculate the estimated mean, minimum, and maximum METHODS population sizes. The estimates of population size in the quadrats (from both meth­ In June of 1997, I marked out 0.5 X 0.5 m quadrats at 2 m ods) were then extrapolated to the entire habitat area by multiplying intervals on twelve algific talus slopes in northeastern Iowa. Each the sample-size estimate by the proportion of the total habitat area slope (population) contained a total of twenty quadrats in which that was sampled. For these calculations, I assumed that the area snails were counted. (The one exception is a large slope, #99, on sampled effectively included the area within the quadrats, between which three distinct algific areas exist. On this slope, I set up ten the quadrats, and a small border around the edge of the transects quadrats in each of the three areas, referred to as B, C, and X. The (up to 0.5 m which is less than half the distance between quadrats) estimates for each area were calculated separately.) when this border contained similar habitat. At each quadrat, I searched for D. macclintocki in the surface litter for three minutes, taking care to minimize the disturbance of the RESULTS underlying loose rocks. Snails found during the search time were marked either with white fingernail polish and an individual number CAPTURE gave high support for Chao's M(th) (Chao 1987) mod­ written in ink or with a uniquely colored and numbered bee tag el (e.g., normalized selection criteria value = 0.83 for both slopes (BeeWorks, Inc.) glued to the shell with superglue. The snails were 33 and 99B) for the populations on which the model selection cri­ then replaced in the quadrat where they were found. teria could be run. The M(th) model assumes the trapping proba­ In order to compare my results with those of a previous study bility varies with time (t) and individual animal (h). Three of the (Wallendorf and Clark 1992) which used sampling boards instead of populations had too few recaptures and the program could not com­ searching quadrats, I also used sampling boards on a subset of slopes. pute the null model results, so the model selection criteria could not Weathered basswood (Tilia americana) boards (0.5 X 0.25 m) were be tested. I decided to use the same model on all populations, and placed adjacent to (< 2 m from) the quadrats. At these points, I the model with the highest uniform support was M(th). This is a marked any individuals which were discovered under the boards. realistic assumption of what is occurring in the snail populations, These snails were readily visible, either attached to the board itself because micro-scale environmental conditions determine whether the or at the surface of the ground underneath, so I did not spend three snails are active at any particular time or location and individuals minutes searching at these points. Marking and handling was iden­ differ in their activity (pers. obs.). tical to the snails found at the quadrats. Chao M(th) estimates of the population sizes of the sampled areas I visited each site a minimum of four times over the summer with range from 59 to 2,333 (Table 1). Gazey and Staley estimates range a minimum of two days between visits. In order to test for indepen­ from 76 to 2,973 (Table 2). Minimum sample sizes as determined dent sampling, I plotted captures according to quadrat location on from Gazey and Staley's Bayesian method are given in Table 2. Es­ semivariograms using the GSLIB Geostatistical Software package timates for each population from both methods (Bayesian and Chao 36 JOUR. IOWA ACAD. SCI. 107(2000)

Table 1. Estimates of the population size in the areas sampled using Chao's M(th) method. These estimates are extrapolated to obtain total population estimates over the entire habitat area shown in Table 3. Population sizes could not be estimated for all slopes with this method due to the small sample sizes and low recapture rates. Slope numbers refer to numbers assigned in original reports by Frest (1981-1987). Slope 99 is divided into three distinct areas B, C, and X which were studied separately. Probability of capture (p) is the probability of capture taken from the average p from all capture occasions calculated by Chao's M(th) model. Confidence intervals (Cl), total number of snails captured (n), number of snails recaptured (r), percentage of individuals recaptured (%r), and number of sampling occasions at that site (s) are also listed.

Slope number Chao's M(th) (SE, 95% Cl) p n r %r s

297 405 (326, 117-1,667) 0.016 32 2 6.25 5 207 59 (45, 24-247) O.Q78 16 1 0 4 103 149 (73, 71-393) 0.047 35 5 14.29 6 247 101 (96, 30-516) 0.043 16 1 6.25 4 120 296 (171, 120-886) 0.032 45 3 6.67 5 99B (total) 367 (63, 272-528) (boards only) 100 (15, 80-144) 0.156 63 30 47.6 8 (quadrats only) 432 (187, 209-1,008) 0.023 69 6 8.70 8 99C (total) 136 (42, 85-263) (boards only) 70 (19, 48-131) 0.105 35 8 22.86 8 (quadrats only) not estimated 14 0 0 8 99X (total) 157 (37, 110-267) (boards only) 90 (19, 68-152) 0.230 56 22 39.29 4 (quadrats only) not estimated 19 0 0 4 62 539 (540, 125-2,829) 0.010 34 1 2.94 6 98 570 (270, 256-1,418) 0.022 78 4 5.13 6 121 (total) 654 (115, 478-937) (boards only) 75 (15, 57-119) 0.127 44 11 25.00 6 (quadrats only) 831 (252, 486-1,521) 0.030 142 11 7.75 6 33 1,874 (411, 1,252-2,901) 0.024 297 23 7.74 7 119 2,333 (964, 1,111-5,152) 0.013 175 6 3.43 6 246 not estimated 47 0 0 5

M(th)) are plotted against one another in Figure 1. Although Bayes­ ian estimates tend to be higher, the slope of the regression line is 1, Table 2. Means, minimum, and maximum population sizes for indicating the methods hold the same relationship over the range of the sampled area calculated using Bayesian method from Gazey population sizes found in this study. and Staley (1986). Slope numbers refer to numbers assigned in The population estimates from these data extrapolated over avail­ original reports by Frest (1981-1987). Slope 99 is divided into able habitat are compared to previous estimates from Frest (1981- three distinct areas B, C, and X which were studied separately. 1985) (Table 3). The sampled areas, along with the total suitable Only snails observed in quadrats were considered for these snail area, are given in Table 3. analyses. These estimates are extrapolated over the entire po­ Overall capture probabilities were much higher for the boards tential habitat area to obtain the estimates in Table 3. than for the quadrats (0.127 vs. 0.03 on slope 121; 0.156 vs. 0.023 on 99B). Snails were also recaptured at much higher rates under the Slope Mean Minimuma Maximum a boards than in the quadrats (25.0%, 47.6%, 39.29%, and 22.86% vs. 7.75%, 8.70%, 0%, and 0%). A comparison of board and quadrat 297 135 78 182 captures for slope 99 is shown in Figure 4. 207 76 28 138 103 100 53 171 Movement 247 107 32 231 120 216 132 281 The snails moved very little during the survey. The percentage of 99B 771 330 1,210 recaptured snails that moved between quadrats was 0.13% on slope 99C 230 30 880 33, 0.09% on slope 121, 0.17% on slope 119, and 0.05% on 99X 99X 1,074 225 1,825 (between boards). The longest recorded movement was 8 meters. 62 704 192 1,292 Table 4 lists the individuals that moved, how far away from their 98 740 347 1,216 original location that they were subsequently found, and their speed. 121 1,378 700 2,500 Low inter-sample movement rates mean that the assumed sampled 33 1,832 1,320 2,500 area is not underestimated, because few snails even moved to the next quadrat. 119 2,787 1,325 5,100 246 2,973 815 4,850 Variability in time and space a These values correspond to the 95% confidence intervals as de- Snails were not evenly distributed on all of the slopes. On some termined visually from graphs. slopes, captures occurred only in a few quadrats, whereas other slopes DISCUS MACCLINTOCKI POPULATION SIZES 37

Comparison of Population Estimate Methods ing studies using boards, where the recapture rate was 54% (Ostlie 1992). The total number of individuals under the boards also ap­ peared to be more consistent than the number in the quadrats (Fig. 4). These behaviors suggest that the snails are more likely to come to the surface under a board than in an open quadrat, and once under a board, they are more likely to remain under a board. The snails' behavior is likely due to the boards maintaining cover over the area and aiding in moisture retention on the ground. In effect, the snails were "trap happy", a classic problem in population estimation. Be­ cause the estimates are based on the proportion of captures which are recaptures, this could severely bias the population estimates to be lower than the actual population size (White et al. 1982).

Comparison with past estimates of Discus macclintocki population size

500 1000 1500 2000 2500 3000 Poputatlon &tlmate U.. ng Chao's M(th) No standard errors are available for Frest's population size esti­ mates, so traditional significance testing is not possible. However, Fig. 1. Relationship between mean population estimates of the sam­ eight of twelve of my estimates are much higher than his estimates pled area obtained from two different statistical methods (Chao's M(th) (Table 3). Using a sign test, my estimates as a group are not signif­ and Gazey and Staley's Bayesian methods). The solid line represents icantly higher than Frest's estimates as a group (p = 0.121). This the relationship between the estimates (y = l.12x + 67.59, r2 = 0.95, p < 0.05). The dashed line represents a 1:1 relationship between the sign test does not test for significant differences between pairs of two methods. The data points correspond to the eleven slopes (or estimates, but rather whether more than an expected number of es­ areas) which have estimates available from both methods using quad­ timates from one method are larger than another method. Thus, rats only (Tables 1 and 2). neither Frest's nor my method can be considered uniformly biased in one direction. Some of the differences can probably be explained by the structure of the particular slope on which the populations exist. For example, showed a more even distribution of captures (Fig. 2). Semivariograms slope 247 consists of a small patch of talus and a large overhanging showed no relationship between number of captures and capture lo­ cliff with very little vegetation, which could not be sampled. The cation (Fig. 3). Figure 4 shows the total captures at each sampling potential snail habitat was calculated only as the talus area and the time on slope 99 where boards and quadrats are treated separately. area at the base of the cliff where many shells and a few live snails Notice the variability in number of captures between sampling were found. However, Discus macclintocki has been found living in times. cracks in bare cliff faces on other slopes (pers. obs.). If the entire cliff face is counted as habitat, the estimate would be greatly increased. DISCUSSION Thus, a large amount of variance in my estimates is probably due to Comparison of population estimates using different models area effects. The assumed sample area (calculated from the area which is cov­ The population estimates calculated in this study show large con­ ered by the quadrats) may be larger than the actual area sampled fidence intervals which is expected given the small sample sizes and because movement is limited (at least at the surface), but, in that low numbers of recaptures. Although CAPTURE estimates tend to case, these population estimates would be underestimates of the true be biased low with small sample sizes (Koper and Brooks 1998), all population size. In addition, with such low rates of movement, it is of the Bayesian estimates fell within the standard errors of the Chao unlikely the assumed area sampled is underestimated. Therefore, the M(th) estimates for populations where both estimates were available assumed area sampled is probably appropriate for minimum popu­ (Table 3 and Fig. 1). Because the Bayesian method does not require lation size estimates. large sample sizes, it was able to estimate population sizes for 99C, Population estimates for four slopes are smaller than those ofFrest, 99X, and 297 where the Chao estimator did not have enough data which is likely due to the large habitat areas given by Frest. Frest's to generate an estimate. area estimate (52,000 m2) is larger than my estimate (300 m2) for Difficulties do exist with the Bayesian methods. As pointed out slope 297 (the only slope where both estimates are available). It is in Chao (1989), estimates calculated using the method of Gazey and likely that Frest was estimating the entire hillside area rather than Staley (1986) can be influenced by the assumed size of the prior just suitable habitat area. In that case, I would have underestimated distribution and is especially sensitive to the length of the tails of the total population size since the area I sampled actually represents the distribution curve. I minimized the tails by running test distri­ a larger proportion of useable habitat than I assumed when extrap­ butions ro find a suitable range as Gazey and Staley suggest (1986) olating my data to the whole slope. ro minimize this problem. Wallendorf and Clark (1992) used sampling boards for their mark­ recapture study on slope 99 (encompassing the three separate areas Cover boards versus quadrat sampling B, C, and X). Their population estimate from CAPTURE of205,000 Overall estimates from the boards and quadrats are not directly (95% c.i. = 47,000 to 885,000) is significantly higher than the comparable because fewer boards were used than quadrats. However, estimate from my study. However, they assumed uniform density sampling boards produced very different dynamics in the snails' be­ over the entire slope area to obtain this value. The assumption of havior and trapability. Capture probabilities were much higher for uniform density is not realistic, as can be seen from the results from the boards than for the quadrats and snails were recaptured at much the estimates from the three distinct sections (areas B, C, and X) higher rates under the boards than in the quadrats. The recapture sampled on slope 99. Extrapolating a high density over areas which rates under boards (23-48%) are consistent with previous monitor- do not contain many (if any) snails would result in a higher estimate. 38 )OUR. IOWA ACAD. SCI. 107(2000)

Table 3. Comparison of total population estimates from this study and those previously reported by Frest (1982, 1985, 1986). The estimates in the first two columns are calculated by extrapolating estimates (Tables 1 and 2) for the sampled area over the total potential habitat area. Confidence intervals (Cl) are also extrapolated from the confidence intervals in tables 1 and 2. Slope numbers refer to numbers assigned in original reports by Frest (1981-1987). Slope 99 is divided into three distinct areas B, C, and X which were studied separately. Slopes marked with - could not be estimated under Chao's M(th) model.

Area Potential Chao M(th) Bayesian Sampled Snail Habitat Population (95% Cl) (95% Cl) Frest's estimates (m2) (m2)

297 810 (234-3,334) 270 (156-364) 4,000-6,000 150 300 207 22,125 (9,000-92,625) 28,500 (10,500-51,750) 2,000 80 30,oooa 103 298 (142-786) 200 (106-342) 2,000 69 138 247 182 (54-929) 193 (58-416) 500 70 126 120 740 (300-2,215) 540 (330-703) 4,000 186 465 99total 18,667 2,000 99B 941 (455-2,196) 1,680 (719-2,636) 157 342 99C 4,262 (556-16,308) 77 1,427 99X 19,690 (4,125-33,458) 36 660 62 981 (228-5' 150) 1,282 (350-2,352) 400-600 78 142 98 1,629 (731-4,051) 2,114 (991-3,474) 1,000 84 240 121 11,946 (6,986-21,864) 19,809 (10,063-35,938) 2,000 72 1,035 33 20,698 (13,828-32,041) 20,234 (14,579-27 ,612) 600-1,000 67 740 119 21,602 (10,287-47,704) 25,806 (12,269-47 ,222) 2,000 108 1,000 246 297,300 (81,500-485,000) 1,000 150 1s,ooob a Slope size estimate from Frest (1983) b Slope size estimate from Frest (1985)

Using their density estimate of 86 individuals per m2 over the total species (Goodhart 1962, Woodruff 1978, Baur 1988, Pfenninger et area I sampled on slope 99 (270 m2), results in a population estimate al. 1996, among others). Although movement is rare in D. macclin­ of 23,220 individuals which is still much larger than the Bayesian tocki, evidence from genetic studies indicates that gene flow within maximum estimates for the three areas (B, C, X) combined (3,915 a slope does occur (Ross 1999). individuals). Capture probabilities varied considerably among slopes (as well as Table 4. Individuals that were recaptured at a quadrat or among the three areas of slope 99) so an estimate of the average board other than where they were initially captured and where density of snails is inappropriate for the entire slope. Population size they moved. Note that these are maximum rates of movement estimates for the sampled area divided by sample area (surface area based on single events. Caution should be used in extrapolating only) give results ranging from 0.74 individuals per m2 on slope these rates to longer time periods. Actually, the rates would be 207 to 28.07 individuals per m2 on slope 33. Wallendorf and Clark much lower if figured as distance moved over the entire time (1992) recorded average densities of Discus macclintocki of 86 individ­ observed and in fact for most individuals in the study, move­ uals per m2. Discus cronkhitei Newcomb, a sister species of D. mac­ ment was zero. clintocki, has been found in densities as high as 48.9 per m2 (Van Es and Boag 1981). Densities of terrestrial snails vary with species and Distance Moved habitat and ranges from 0.1-1.5 Cepaea nemoralis per m2 (Greenwood Between Cap- Rate/meters 1974) to 794 Chondrina clienta per m2 (Baur 1988) and everywhere Individual ture/meters per day in between (Goodhart 1962, Woodruff 1978, Cowie 1984, Baur and Baur 1990, Pfenninger et al. 1996). Due to the large range in snail Slope 33 densities, it is impossible to conclude whether D. macclintocki is found at a lower density than expected. #97 2.8 0.029 #60 8.0 0.086 #112 Movement 2.0 0.667 Slope 121 Very few snails (less than 0.2% on any particular slope) moved to #53 2.0 a different quadrat or board during this study. This low movement 0.667 probability is typical for snails; for example, Schilthuizen and Lom­ Slope 119 baerts (1994) found that 90% of adult Albinaria corrugata, which #53 2.0 0.667 also resides in a patchy habitat, moved less than 2 m. Of the Discus Slope 99X macclintocki that moved in this study, the most common rate of move­ ment was 0. 7 m per day. These rates are similar to the rates of #R32 movement a 2.0 0.667 movement in a study by Ostlie (1992) in which Discus macclintocki movement b 2.0 0.667 moved up to 1 m per week which is not unusual for other snail #R80 2.0 0.667 DISCUS MACCLINTOCKI POPULATION SIZES 39 -, (a) Slope 103 (A) Slope 98 "O E :I I 'C. 6 CJ G) "' CJ .!! ·;; :n c: ca c: I I I I I ~ I I I I I I I l"""'I ·;:: 4 Vl ,... I ,... <( c::; ca < "' "'<( <( "'<( "'CJ CJ"' CJ CJ"' > Quadrat" Number ·e 2 G) Cl) 0 0 2 4 6 (bl Slope 119 Lag Distance, meters

,... (B) Slope 33 "'<( "'<( <( "'<( Ill"' "'Ill Quadrats G) 100 CJ c: 80 ca ·;:: 60 ca ·e> 40 (Cl Slope 120 ------G) 20 Cl) 0 0 2 4 6 H' ,I,., ,1.1,I .• ,., Lag Distance, meters ,... "'<( "'<( Ill"' Ill"' Ill Ill"' Quad rats (C) Slope246 Fig. 2. Histograms of number of initial captures at each different quadrat on slopes 103, 119, and 120. Quadrat numbers refer to the G) 25 identification of the particular quadrat and are not necessarily arranged CJ c: 20 linearly, so quadrats adjacent to one another on the graph are not ca ·;:: 15 necessarily adjacent in space. Some slopes showed a clumped distri­ ca bution of captures (A), while others showed an even distribution (B, ·e> 10 C). For histograms of all of the slopes, see Ross (1998). G) 5 Cl) 0 Spatial and temporal variation 0 2 4 6 Snails are not uniformly distributed across algific talus slopes. On Lag Distance, meters some slopes most of the initial captures are from just a few sampling sites (Fig. 2). This may reflect an aggregation of snails into colonies Fig. 3. Semivariograms showing the relationship of the number of as suggested by Frest (1984), or the snails may only be more uni­ captures at each quadrat compared to the number captured at quadrats formly distributed beneath the surface and are funneled to the surface at the lag distances shown. Most slopes showed patterns suggesting at certain places, i.e. those areas cooled just enough by air flow from there was no spatial relationship (A, B). A few slopes seem to show a vents or with sufficient vegetation to maintain desirable humidity slight spatial relationship (C). For semivariograms of all of the slopes, levels. see Ross (1998). Semivariograms show no consistent relationship between number of captures and capture location (Fig. 3). Semivariograms measure the relationship between the variation (semivariance) of some item some slopes captures are concentrated in smaller areas than the entire of interest (in this case the number of initial captures at one point, area sampled. No single pattern is seen on all of the slopes, rein­ and the same item of interest at points at increasing distances, forcing the idea that each slope has a unique arrangement of vents termed lag distances, from that point. Lag distances ranged from and desirable vegetation. two meters (the distance between the quadrats in this study) to six I sampled movement in a horizontal plane (parallel to the surface), meters, for this analysis. If samples were spatially non-independent, however vertical movement could also affect the distribution of the the semivariograms would show a slowly increasing slope. Instead, snails. Locasciulli and Boag (1987) found snails in forest litter tended most semivariograms show rapid increases before leveling off (Fig. to move upward in the summer. Boag (1985) found that humidity 3A, B), which means there is no relationship. Therefore, the indi­ level is the most important factor regulating the presence of snails vidual quadrats can be considered independent. above the litter. However, in that study, 67% of the Discus cronkhitei Semivariograms from three slopes (246, 103, and one transect of (a close relative of Discus macclintocki) were found deeper than 5 cm 297) seem to show a slight spatial relationship, indicating that on from the surface. If only a small portion of the snail population is 40 JOUR. IOWA ACAD. SCI. 107(2000)

as is evident in the following anecdotal example. In May and June (A) Slope 99, Area B of 1996, I observed ten Discus macclintocki on slope 99 marked in a "O.... 'i 1991 study where 122 adults were marked with fingernail polish in unique color patterns (Ostlie 1992, Wallendorf and Clark 1992). ~ a 3200 I .Quad rats " Jl _n D Boards However, during 1997, I observed only one snail from that study, ~u~ ~.~, 1 ~ 1 ~ 1 E1E1~1 - .!l and it was not one of the individuals observed in 1996. Long-term .g ·: 25- 28- 01-Jul 04-Jul 09-Jul 12-Jul 15-Jul 20- ~ oo. Jun Jun Sep monitoring is necessary to determine just how much snail popula­ Date tions fluctuate. Therefore, it is difficult to attribute larger population estimates in this study to actual population increases.

Conclusion Snails live at much finer spatial scales than what is generally rec­ (B) Slope 99, Area C ognized by researchers. Spatial heterogeneity in microhabitat con­ ditions can have a large effect on whether a snail is present or absent in one area or another, even within a slope. Although assuming 5g.10~ l 15I habitat uniformity even within "snail habitat" tends to overestimate n the population, it is clear that sizeable populations exist at all of i~: Jl u,Jl,JJ,ll, o,JJ,JJ, these sites. Estimates presented here are probably biased low due to .... "' 25-Jun 28-Juu OJ-Jul 04-Jul 09-Jul 12-Jul 15-Jul 20-Sep error in estimates of potential habitat, trap-happy snails, and the Date inability to monitor snails underground. Population sizes appear much larger at most sites than previously thought, although it is not clear whether this is due to bias in previous sampling methods or actual increases in populations since the previous estimates were made. Despite the difficulties in sampling, this study provides valid (C) Slope 99, Area X minimum estimates which will serve as baseline data for monitoring efforts and future studies of Discus macclintocki. Surveys with wider coverage over a longer time period may result in more precise pop­ ulation estimates. Future monitoring should be conducted with con­ JJ, sistent area estimates and sampling methodology such as that out­ Jl,JJ lined in this study, in order to make comparisons with these data. 12-Jul 15-Jul 20-Sep Date ACKNOWLEDGEMENTS I would like to thank Tawnya Cary, Katherine Holger, Tamra Fig. 4. Plots of the total number of individuals captured at each sam­ Danielson, Jessemine Fung, and Adam Remsen for their help in the pling occasion on slope 99. Snails caught under boards and those field. Funding for this project was provided by the Iowa Department found in quadrats are displayed separately. of Natural Resources. I appreciate the cooperation of the Iowa De­ partment of Natural Resources (especially Daryl Howell), the Algific Slope National Refuge, the U.S. Fish and Wildlife Service, and the active at the surface at any one time and only at desirable locations, Iowa Chapter of The Nature Conservancy. This study was conducted then the boards and quadrats could be measuring only a very small under the Iowa Department of Natural Resources Permit for research fraction of the actual population. If this is the case, the population on Discus macclintocki. The support and assistance of my major pro­ estimates given here could be severe underestimates of the actual fessors, Brent Danielson and Richard Hoffmann, is greatly appreci­ populations. ated. Temporal variation in activity levels is also extremely important in these populations. Both day-to-day and year-to-year fluctuations LITERATURE CITED in presence/absence are evident. Many factors could influence this AUSDEN, M.A. 1996. Invertebrates. Pages 139-177. In Ecological Census result including moisture levels and temperature over the season. In Techniques: A Handbook. W. ). Sutherland, ed. Cambridge University a study of habitat variables associated with the presence of D. mac­ Press, Cambridge. clintocki at a board, Ostlie (1992) showed the number of individuals BAKER, F. C. 1928. Description of new varieties of land and freshwater at each board varied over time and with temperature, but the peaks mollusks from Pleistocene deposits in Illinois. Nautilus 41:132-137. BAUR, B. 1988. Microgeographical variation in shell size of the land snail of activity were not the same over the entire slope. Although, Wal­ Chondrina c!ienta. Biological Journal of the Linnean Society 35:247-259. lendorf and Clark (1992) found no difference in the number of snails BAUR, A. and B. BAUR. 1990. Are roads barriers to dispersal in the land captured at temperatures between 6 and 12°C, these temperatures snail Arianta arbustorum? Canadian Journal of Zoology 68:613-617. are below the temperature ( 15 .6°C) at which Ostlie ( 1992) noted BOAG, D. A. 1982. Overcoming sampling bias in studies of terrestrial reduced activity. The patterns of variation in activity are not consis­ gastropods. Canadian Journal of Zoology 60:1289-1292. tent among slopes, further emphasizing the uniqueness of each slope. BOAG, D. A. 1985. Microdistribution of three genera of small terrestrial Year-to-year variation in population size was formerly attributed snails (Stylommatophora: Pulmonata). Canadian Journal of Zoology 63: to population decline. For example, Frest (1984) compared popula­ 1089-1095. tion sizes between years at five slopes and found three of them to be CAIN, A.). and). D. CURREY 1968. Studies on Cepaea. III. Ecolgenetics of a population of Cepaea nemora!is (1.) subject to strong area effects. lower in the second year. However, fluctuations between sampling Philosophical Transactions of the Royal Society of London, Series B, 253: occasions in this study suggest year-to-year variation could be a nor­ 447-482. maI occurrence if sampling is not done on multiple days each year. CHAO, A. 1987. Estimating the population size for capture-recapture data Even individual snails may vary in their activity from year to year, with unequal catchability. Biometrics 43:783-791. DISCUS MACCLINTOCKI POPULATION SIZES 41

CHAO, A. 1989. Estimating population size for sparse data in capture­ (Discus macclintocki) 1992 field season. Report to the Iowa Chapter of The recapture experiments. Biometrics 45:427-438. Nature Conservancy, Des Moines. COWIE, R. H. 1984. Density, dispersal and neighbourhood size in the land OTIS, D. L., K. P. BURNHAM, G. C. WHITE and D. R. ANDERSON. snail Theba pisana. Heredity 52:391-401. 1978. Statistical inference from capture data on closed animal popula­ DEUTSCH, C. V. and A. G. JOURNEL. 1992. GSLIB Geostatistical soft­ tions. Wildlife Monographs 62:1-135. ware library and user's guide. Oxford University Press, New York. PFENNINGER, M., A. BAHL and B. STREIT. 1996. Isolation by distance FREST, T. J. 1981. Project SE-1-2. Iowa Pleistocene snail. Report to the in a population of a small land snail Trochoidea geyeri: evidence from direct Iowa State Conservation Commission, Des Moines. and indirect methods. Proceedings of the Royal Society of London, Series FREST, T. J. 1982. Project SE-1-4. Iowa Pleistocene snail. Report to the B, 263:1211-1217. Iowa State Conservation Commission, Des Moines. PILSBRY, H. A. 1948. Land mollusca of North America (north of Mexico), FREST, T. J. 1984. National Recovery Plan for Iowa Pleistocene snail. Unit­ Volume II, Part 2. Academy of Natural Science of Philadelphia Mono­ ed States Fish and Wildlife Service. graphs No. 3. FREST, T. J. 1985a. Final report Iowa Pleistocene snail project 1983. Report ROSS, T. K. 1998. Gene flow at a snail's pace: phylogeography and conser­ to the Iowa State Conservation Commission, Des Moines. vation genetics of relict populations of the Iowa Pleistocene snail. Ph.D. Dissertation, Iowa State University. FREST, T. J. 1985b. Final report Iowa Pleistocene snail survey 1985. Report ROSS, T. K. 1999. Phylogeography and conservation genetics of the Iowa to the Iowa State Conservation Commission, Des Moines. Pleistocene snail. Molecular Ecology 8:1363-1374. FREST, T. 1986. Iowa Pleistocene snail survey. Project E-1-6. Report to J. SCHILTHUIZEN, M. and M. LOMBAERTS. 1994. Population structure the Iowa Department of Natural Resources, Des Moines. and levels of gene flow in the Mediterranean land snail Albinaria corrugata FREST, T. J. 1987. Final report. Iowa Pleistocene snail project, 1987. Project (Pulmonata: Clausiliidae). Evolution 48:577-586. E-1-7. Report to the Iowa Department of Natural Resources, Des Moines. SOLEM, A. 1976. Final Report, Contract no. 14-16-0008-965. U. S. De­ GAZEY, W. J. and M. J. STALEY. 1986. Population estimation from mark­ partment of the Interior, Office of Endangered Species. recapture experiments using a sequential Bayes algorithm. Ecology 67: STATE OF IOWA. 1988. Iowa Administrative Code. Section 571, Chapter 941-951. 77.1. GOODHART, C. B. 1962. Variation in a colony of the snail Cepaea nemoralis TOMIYAMA, K. and M. NAKANE. 1993. Dispersal patterns of the Giant (L.). Journal of Animal Ecology 31:207-237. African Snail, Achatina fulica (Ferussac) (Stylommatophora: Achatinidae), GREENWOOD, J. J. D. 1974. Effective population numbers in the snail equipped with a radio-transmitter. Journal of Molluscan Studies 59:315- Cepaea nemoralis. Evolution 28:513-526. 322. HUBRICHT, L. 1955. Discus macclintocki. Nautilus 69:34. U.S. FISH AND WILDLIFE SERVICE. 1993. Endangered and threatened KOPER, N. and R. J. BROOKS. 1998. Population-size estimators and un­ wildlife and plants. United States Fish and Wildlife Service, Department equal catchability in painted turtles. Canadian Journal of Zoology 76: of the Interior. 458-465. VAN ES, J. and D. A. BOAG. 1981. Terrestrial molluscs of Central Alberta. KUZNIK, E. 1997. Comparison of two methods of quantitative sampling Canadian Field-Naturalist 95:75-79. applied in studies on land malacocenoses (Gastropoda: Pulmonata). Ma­ WALLENDORF, M. J. and W.R. CLARK. 1992. Evaluation of a population lakologische Abhhandlungen Staatliches Museum fur Tierkunde Dresden monitoring methodology for Discus macclintocki: 1992 field season. Report to the Iowa Chapter of The Nature Conservancy. 18:263-270. WHITE, G. C., K. P. BURNHAM, D. L. OTIS and D. R. ANDERSON. LEONARD, A. B. 1959. Handbook of gastropods in Kansas. University of 1978. User's Manual for Program CAPTURE Utah State University Kansas Museum of Natural History, Publication 20. Press, Logan, Utah. LOCASCIULLI, 0. and D. A. BOAG. 1987. Microdistribution of terrestrial WHITE, G. C., D.R. ANDERSON, K. P. BURNHAM and D. L. OTIS. snails (Stylommatophora) in forest litter. Canadian Field-Naturalist 101: 1982. Capture-Recapture and Removal Methods for Sampling Closed 76-81. Populations. Los Alamos National Laboratory, Los Alamos, New Mexico. OGGIER, P., S. ZSCHOKKE and B. BAUR. 1998. A comparison of three U.S. Department of Energy. methods for assessing the gastropod community in dry grasslands. Pe­ WILLIAMSON, P., R. A. D. CAMERON and M. A. CARTER. 1977. Pop­ dobiologia 42:348-357. ulation dynamics of the landsnail Cepaea nemoralis L.: a six-year study. OSTIIE, W.R. 1992. Development of a monitoring methodology for Discus Journal of Animal Ecology 46:181-194. macclintocki, 1991 field season (with notes on the life history of the spe­ WOODRUFF, D. S. 1978. Evolution and adaptive radiation of Cerion: a cies). Report to the Iowa Department of Natural Resources, Des Moines. remarkably diverse group of West Indian land snails. Malacologia 17: OSTIIE, W. R. 1993. A monitoring method for the Iowa Pleistocene snail 223-239.